Submission to Ieee Transactions on Software Engineering a Controlled Experiment for Validating the Personal Software Process Methodology

The Personal Software Process is a process improvement methodology aiming at individual software engineers. It claims to improve software quality (in particular defect content), effort estimation capability, and process adaptation and improvement capabilities. We have tested some of these claims in a controlled experiment comparing the performance of participants who had just previously received a PSP course to a control group of participants who had received other technical training instead. Each participant of both groups performed the same task. The results indicate that the PSP group estimated their productivity more accurately , produced more reliable programs, and showed smaller variability in performance than the control group. We conclude that PSP training reduces the risk in a software development eeort, although the improvements are smaller than some PSP proponents claim. 1 The Personal Software Process (PSP) methodology The Personal Software Process (PSP) methodology for improving the software process was introduced in 1995 by Watts Humphrey 4]. PSP is an application of the principles of the Capability Maturity Model (CMM, 3]) on the level of an individual software engineer. In contrast to the CMM, however, which allows only for assessment of process quality, the PSP makes concrete methodological and learning suggestions, down to the level of a 15-week course with rather speciic procedural content. The goals of the PSP are that an individual software engineer learns how to accurately estimate, plan, track, and re-plan the time required for individual software development eeorts, how to work according to a well-deened process, how to deene and reene the process, how to use reviews eeectively and ee-ciently for improving software quality and productivity (by nding defects early), how to avoid defects, how to analyze measurement data for improving estimation, defect removal, and defect prevention, how to identify and tackle other kinds of process deeciencies. The main basic techniques used are gathering objective measurement data on many aspects 1